纹理图像分析技术在表面粗糙度测量中的应用

摘要

以MATLAB软件为设计平台,构建了零件表面粗糙度的检测系统.采用同态滤波等图像处理技术,对零件表面图像进行预处理,选取图像灰度直方图参数为表征纹理的像元特征值,灰度共生矩阵参数为表征纹理结构的区域特征值,建立了BP神经网络,实现了对零件表面粗糙度参数的测量,并对测试结果与电动轮廓仪的测量结果进行了对比和分析,验证了系统的可行性.%Taking MATLAB as the platform of software design,a surface roughness measurement system was developed.By using image analysis technologies such as homomorphic filter to preprocess the surface image of parts,and selecting the parameters of image gray-level histogram as the picture element eigenvalue of texture and the parameters of graylevel co-occurrence matrix as the area eigenvalue of texture structure,a BP neural network was set up,and the parameters of surface roughness of parts were measured.The results of test were compared with those of the electric contour graph measurement and were analyzed to verify the feasibility of this measurement system.